Abstract
Background
Anemia is encountered in patients with nonvalvular atrial fibrillation (NVAF) on oral anticoagulants (OACs) but the prognostic impact was not well scrutinized in real‐world settings.
Methods
We conducted a historical multicenter registry of patients with NVAF taking OACs at 71 centers in Japan. Those with mechanical heart valves or a history of pulmonary thrombosis or deep venous thrombosis were excluded. Anemic patients were divided into three groups of hemoglobin (Hb) level: moderate/severe (Hb < 11.0 g/dL), mild (men: Hb of 11.0–12.9 g/dL; women: Hb of 11.0–11.9 g/dL), and no anemia. The endpoints included major bleeding, hemorrhagic stroke, ischemic events, ischemic stroke, and all‐cause mortality.
Results
Among 7558 consecutive patients (mean age, 73 years; men 67%) registered in February 2013 and followed until February 2017, 2100 (28%) patients had anemia. The anemic patients were older (moderate/severe: 79 years; mild: 77 years; no anemia: 71 years, p < .001), and HAS‐BLED score was significantly higher in the anemic patients (p < .001). The cumulative incidences at 4 years of major bleeding in moderate/severe, mild, and no anemia group, were 14.9%, 10.7%, and 6.4%, respectively. The adjusted hazard ratios (HRs) (95% confidential intervals (CIs)) of major bleeding of moderate/severe and mild anemia group were 1.96 (1.49–2.58) and 1.48 (1.17–1.87) compared to no anemia group. The adjusted HRs (95% CIs) for ischemic events were 0.63 (0.39–0.99) and 1.03 (0.76–1.39).
Conclusions
The severity of anemia in the patients with NVAF on OACs was associated with major bleeding.
Keywords: anemia, atrial fibrillation, bleeding, oral anticoagulants, registry
Anemia was frequently encountered in patients with NVAF on OACs in daily clinical practice and significantly associated with the incidence of major bleeding and all‐cause mortality, but not with the incidence of ischemic events or strokes. Anemic status should be carefully monitored for such patients.
1. INTRODUCTION
Anticoagulation is the standard of care for patients with thrombotic risk due to nonvalvular atrial fibrillation (NVAF), which is inevitably associated with the risk of hemorrhagic complications. 1 We thus need to pay attention to patients with high bleeding risks. The commonly used bleeding risk score for such patients is the HAS‐BLED score, which consists of hypertension, history of stroke, renal or liver dysfunction, history of major bleeding, labile anticoagulation control, elderly patients (>65 years), and concomitant use of drugs or alcohol. 2 The HAS‐BLED score is well established and routinely used in clinical practice, but the definition of each component should be investigated. One of the components of HAS‐BLED is B, which was defined as a history of bleeding events or predisposition of bleeding. However, there are no reports, including original studies, that defined anemia as a bleeding predisposition.
Patients with NVAF were generally senile, and the mean age was reported as 75 years in Asia and 70 years in the United States from the claim database. 3 , 4 As a result, these patients had many comorbidities or abnormalities in laboratories when oral anticoagulants (OACs) were administered. Although anemia is one of the commonly encountered conditions of the elderly in daily clinical practice, the data on the prevalence and effects of anemia in patients with NVAF in daily clinical practice are scant. 5
In addition, randomized clinical trials of OACs generally excluded those with severe anemia (hemoglobin (Hb) < 10 g/dL in RE‐LY trial; Hb <9 g/dL in ARISTOTLE trial). 6 , 7 Even without evidence of efficacy and safety of OACs in such populations, as well as the prevalence of severe anemia, physicians or patients should consider the use of OACs in conjunction with the current risk stratification on HAS‐BLED score for anemia. Thus, it should be helpful in clinical practice to estimate the prevalence of anemic patients among those with NVAF on OACs and compare the bleeding risks according to the degree of anemia. We thus analyzed a large historical registry of consecutive patients with NVAF taking OACs in daily clinical practice.
2. METHODS
We conducted a historical registry study at 71 centers in Japan. We enrolled patients with NVAF taking OACs on February 26, 2013, and followed them until February 25, 2017. We included patients who received a vitamin K antagonist on February 26, 2013, and excluded those with mechanical heart valves or a history of pulmonary thrombosis or deep vein thrombosis (DVT). The institutional review boards of all 71 participating centers approved the study in accordance with the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan and waived the need for written informed consent.
2.1. Data collection and definitions
A review of hospital charts provided the clinical information at baseline, 1, 2, 3, and 4 years of follow‐up. Contacting patients or referring physicians was utilized if necessary. We collected data on patient characteristics, laboratory data, risk of ischemic strokes (CHADS2 score), and risk of hemorrhagic stroke (HAS‐BLED score) at baseline. Definitions of baseline characteristics were described previously. 8 The CHADS2 score was developed to estimate stroke risks in patients with NVAF, 9 and the HAS‐BLED score was developed to predict major bleeding in anticoagulation for such patients. 2
Baseline Hb at the inclusion of each cohort was used. World Health Organization (WHO) defined anemia as an Hb level < 13.0 g/dL for men and < 12.0 g/dL for women. 10 According to the WHO definition, we categorized all patients into three groups: moderate/severe anemia (Hb level < 11.0 g/dL both for men and for women); mild anemia (Hb level of 11.0–12.9 g/dL for men and 11.0–11.9 g/dL for women), and no anemia (Hb level ≥ 13.0 g/dL for men and ≥12.0 g/dL for women). The patients without baseline Hb data were excluded from this analysis.
2.2. Endpoints
The primary endpoint was major bleeding. Major bleeding was defined by the International Society on Thrombosis and Hemostasis criteria. 11 Other endpoints included hemorrhagic stroke, including intracerebral hemorrhage and subarachnoid hemorrhage, ischemic events, ischemic stroke including transient ischemic attack, and all‐cause mortality. Intracerebral hemorrhage was defined as a high‐density area on computed tomography (CT) or a hyperintensity area on magnetic resonance imaging (MRI) T1‐weighted images, indicating bleeding in the brain parenchyma. 12 Subarachnoid hemorrhage was defined as a high‐density area on CT or a hyperintensity area on MRI with fluid‐attenuated inversion recovery in the subarachnoid space. Ischemic events were defined as ischemic stroke, including transient ischemic attack and systemic embolism. 13
2.3. Statistical analysis
We compared patient characteristics and endpoints among the moderate/severe anemia, mild anemia, and no anemia groups. Continuous variables are presented as mean and standard deviation or median and interquartile range, and categorical variables are numbers and percentages. Continuous variables were compared using ANOVA or the Kruskal–Wallis test based on their distribution and categorical variables using the chi‐square test. All patients were censored on the day of the first major bleeding, ischemic events, death, or the last visit if patients were not followed up until the 4‐year study period. We estimated the incidence of endpoints with cases per 1000 patient‐years stratified by the groups. The cumulative incidence was estimated using the Kaplan–Meier method, while intergroup differences were assessed using the log‐rank test.
The risks of moderate/severe anemia or mild anemia relative to no anemia for endpoints were estimated using Cox proportional hazard models and expressed as hazard ratios (HRs) with 95% confidence intervals (CIs). We simultaneously included the variables of moderate/severe anemia and mild anemia as the dummy variables in the models, and no anemia was considered as a reference. We adjusted the clinically relevant variables to estimate the adjusted HRs in the multivariable Cox proportional hazard models. The adjusters for hemorrhagic events were HAS‐BLED score, history of coronary artery disease (CAD), and statin use. The adjusters for ischemic events were CHADS2 score, CAD, history of major bleeding, estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m2 or hemodialysis (HD), and statin use. The adjustment for all‐cause mortality was a combination of these factors. As exploratory analyses, we divided patients in the moderate/severe anemia group into moderate anemia and severe anemia separately and constructed the same Cox proportional hazard models. Because the number of patients with severe anemia was small, these analyses should be considered as exploratory. We also conducted sensitivity analyses by dividing the total cohort into cohort with warfarin only (warfarin alone stratum) and cohort switched to DOACs (switch to DOACs stratum) during the study period and constructed the same Cox proportional hazard models for major bleeding and all‐cause mortality. The missing values for the binary variables were considered “normal” because data should have been available if abnormalities were suspected. Platelet measurements were missed in 19 patients and were not imputed in this study. There were no further missing data on continuous variables except for Hb and platelets.
We constructed the same multivariable Cox proportional hazard models for the primary endpoint in the subgroups to estimate the adjusted HRs for each subgroup and the interaction p values. The subgroups included age (≥75 or <75 years), sex, history of stroke, history of chronic liver disease, history of malignancy, eGFR (≥60, <60 but ≥30, <30 or HD), CHADS2 score (≥2 or <2), HAS‐BLED score (≥3 or <3), and use of antiplatelets.
All statistical analyses were conducted using JMP 15.2 (SAS Institute Inc.). All reported p values were two‐tailed, and p values <.05 were considered statistically significant.
3. RESULTS
3.1. Patient characteristics
We initially registered 8366 patients; of them, 560 were excluded due to conflicting eligibility criteria (n = 16), unclear information in medical records with OAC use (n = 519), or erroneous data (n = 5). In addition, 268 patients without hemoglobin data were excluded from this study (Figure 1). The mean age of the 7558 included patients was 73 years, and 5082 (67%) were men. A total of 2100 (28%) patients had moderate/severe or mild anemia. (Table 1).
FIGURE 1.
Study flowchart. Hb, hemoglobin; NVAF, nonvalvular atrial fibrillation; OACs, oral anticoagulants.
TABLE 1.
Patients' characteristics.
| Total (n = 7558) | Moderate/Severe anemia (n = 726) | Mild anemia (n = 1374) | No anemia (n = 5458) | p value | |
|---|---|---|---|---|---|
| Age‐years, mean (SD) | 73 (10) | 79 (9) | 77 (9) | 71 (10) | <.001 |
| Age ≥ 75 years, n (%) | 3650 (48) | 540 (74) | 924 (67) | 2186 (40) | <.001 |
| Male, n (%) | 5082 (67) | 356 (49) | 950 (69) | 3776 (69) | <.001 |
| Systolic BP‐mmHg, median [IQR] | 126 [116–137] | 124 [112–135] | 126 [113–138] | 127 [117–137] | <.001 |
| Heart rate‐beats/min, median [IQR] | 72 [64–83] | 72 [64–81] | 71 [63–82] | 73 [65–84] | .0003 |
| BMI‐kg/m2, median [IQR] | 24 [22–27] | 22 [20–25] | 23 [21–26] | 24 [22–27] | <.001 |
| BMI < 25.0, n (%) | 4071 (61) | 501 (75) | 860 (69) | 2710 (56) | <.001 |
| TTR‐%, median [IQR] | 84 [39–100] | 71 [28–100] | 79 [34–100] | 86 [43–100] | <.001 |
| Type of AF | |||||
| Paroxysmal, n (%) | 2532 (34) | 253 (35) | 507 (37) | 1734 (33) | .02 |
| Persistent or permanent, n (%) | 4065 (54) | 378 (52) | 693 (50) | 2931 (55) | |
| Unknown, n (%) | 961 (13) | 95 (13) | 174 (13) | 664 (12) | |
| History of stroke, n (%) | 1854 (25) | 208 (29) | 383 (28) | 1263 (23) | <.001 |
| Hemorrhagic stroke, n (%) | 146 (2) | 15 (2) | 26 (2) | 105 (2) | .96 |
| Ischemic stroke, n (%) | 1744 (23) | 199 (27) | 363 (26) | 1182 (22) | <.001 |
| History of major bleeding, n (%) | 271 (4) | 50 (7) | 58 (4) | 163 (3) | <.001 |
| Hypertension, n (%) | 5916 (78) | 594 (82) | 1109 (81) | 4213 (77) | .0009 |
| Diabetes, n (%) | 2362 (31) | 274 (38) | 448 (33) | 1640 (30) | <.001 |
| Insulin therapy, n (%) | 188 (2) | 34 (5) | 42 (3) | 112 (2) | <.001 |
| Heart failure, n (%) | 3224 (43) | 446 (62) | 675 (50) | 2103 (39) | <.001 |
| History of coronary artery disease, n (%) | 2030 (27) | 248 (34) | 445 (33) | 1337 (25) | <.001 |
| ACS, n (%) | 790 (10) | 106 (15) | 179 (13) | 505 (9) | <.001 |
| Peripheral vascular disease, n (%) | 627 (8) | 86 (12) | 145 (11) | 406 (7) | <.001 |
| COPD, n (%) | 324 (4) | 36 (5) | 64 (5) | 224 (4) | .4 |
| Chronic liver disease, n (%) | 584 (8) | 57 (8) | 112 (8) | 415 (8) | .8 |
| Malignancy, n (%) | 856 (11) | 135 (19) | 243 (18) | 478 (9) | <.001 |
| Current smoker, n (%) | 2298 (30) | 187 (26) | 415 (30) | 1696 (31) | .014 |
| CHADS2 score, median [IQR] | 2 [1–3] | 3 [2–3] | 3 [2–3] | 2 [1–3] | <.001 |
| ≥2, n (%) | 5606 (74) | 648 (89) | 1162 (85) | 3796 (70) | <.001 |
| HAS‐BLED score, median [IQR] | 2 [1–3] | 2 [2–3] | 2 [1–3] | 2 [1–2] | <.001 |
| ≥3, n (%) | 1973 (26) | 311 (43) | 471 (34) | 1191 (22) | <.001 |
| B score, n (%) | 271 (4) | 50 (7) | 58 (4) | 163 (3) | <.001 |
| eGFR‐mL/min/1.73 m2, median [IQR] | 59 [47–71] | 40 [27–57] | 51 [40–64] | 62 [51–73] | <.001 |
| 60 ≤ eGFR, n (%) | 3587 (47) | 151 (21) | 458 (33) | 2978 (55) | <.001 |
| 30 ≤ eGFR<60, n (%) | 3465 (46) | 356 (49) | 758 (55) | 2351 (43) | |
| eGFR<30 or Hemodialysis, n (%) | 506 (7) | 219 (30) | 158 (12) | 129 (2) | |
| Platelets‐ × 104/μL, median [IQR] | 180 [151–215] | 181 [141–230] | 176 [143–214] | 181 [154–214] | .03 |
| Platelets<100 × 109/L, n (%) | 234 (3) | 50 (7) | 61 (4) | 123 (2) | <.001 |
| Switch to DOACs, n (%) | 2097 (28) | 149 (21) | 345 (25) | 1603 (29) | <.001 |
| Antiplatelet therapy, n (%) | 1954 (26) | 245 (34) | 445 (32) | 1264 (23) | <.001 |
| Aspirin, n (%) | 1652 (22) | 210 (29) | 355 (26) | 1087 (20) | <.001 |
| Clopidogrel or Prasgrel, n (%) | 340 (5) | 52 (7) | 86 (6) | 202 (4) | <.001 |
| Cilostazol, n (%) | 158 (2) | 27 (4) | 49 (4) | 82 (2) | <.001 |
| Statin, n (%) | 2518 (33) | 174 (24) | 433 (32) | 1911 (35) | <.001 |
| Beta‐blocker, n (%) | 3246 (43) | 306 (42) | 617 (45) | 2323 (43) | .3 |
| ACE‐I/ARB, n (%) | 3947 (52) | 406 (56) | 734 (53) | 2807 (51) | .046 |
| Calcium channel blocker, n (%) | 3498 (46) | 324 (45) | 634 (46) | 2540 (47) | .6 |
| PPI or H2‐blocker, n (%) | 3377 (45) | 438 (60) | 729 (53) | 2210 (41) | <.001 |
| NSAIDs, n (%) | 305 (4) | 44 (6) | 51 (4) | 205 (4) | .01 |
Abbreviations: ACE‐I, angiotensin‐converting enzyme inhibitors; ACS, acute coronary syndrome; AF, atrial fibrillation; ARB, angiotensin II receptor blockers; BMI, body mass index; BP, blood pressure, COPD, chronic obstructive pulmonary disease, DOACs, direct oral anticoagulants; eGFR, estimated glomerular filtration rate; IQR, interquartile range; NSAIDs, nonsteroidal anti‐inflammatory drugs; PPI, proton‐pump inhibitor; SD, standard deviation; TTR, time in therapeutic range.
The mean age was significantly different among the three groups, and the patients with anemia were older than those without anemia (moderate/severe: 79 years; mild: 77 years; no: 71 years, p < .001) (Table 1). There were significantly fewer men in the moderate/severe anemia group than other groups (moderate/severe: 49%; mild: 69%; no: 69%, p < .001). The more anemic, the higher number of patients who had a BMI of <25 among the three groups (moderate/severe: 75%; mild: 69%; no: 56%, p < .001). Anemic patients were more likely to have a history of ischemic stroke, major bleeding, hypertension, diabetes, heart failure, CAD, and malignancy (Table 1). The anemic patients were significantly less likely to be current smokers (6% vs. 9% vs. 15%, p < .001).
In the moderate/severe and mild anemia groups, the CHADS2 score was significantly higher than in the no anemia group (p < .001), as was the HAS‐BLED score (p < .001; Table 1). The prevalence of B score was low even in the moderate/severe anemia group (50/726, 7%). Among 7558 patients on warfarin at baseline, 2097 (28%) switched to DOACs during the follow‐up period. Those switched to DOACs were less common in the moderate/severe anemia group than in other groups (moderate/severe: 21%; mild: 25%; no: 29%, p < .001).
3.2. Endpoints
At 4 years, 59% of patients completed the follow‐up, and the median follow‐up was 3.9 years. The cumulative incidences at 4 years of major bleeding in the moderate/severe, mild, and no anemia groups were 14.9%, 10.7%, and 6.4%, respectively (Figure 2A). The incidence of major bleeding with cases per 1000 patient‐years among these three groups was 45.2 versus 28.6 versus 16.0, respectively (Table 2). The adjusted HRs (95% CIs) of moderate/severe and mild anemia for major bleeding were 1.96 (1.49–2.58) and 1.48 (1.17–1.87) compared to the no anemia group (Table 2). The cumulative incidences at 4 years of hemorrhagic stroke in the moderate/severe, mild, and no anemia groups were 0.8%, 1.1%, and 1.0%, respectively (Figure 2B). The incidence of hemorrhagic stroke with cases per 1000 patient‐years in these three groups was 2.8 versus 2.6 versus 2.5, respectively (Table 2).
FIGURE 2.
Cumulative incidence of endpoints. (A) Major bleeding. (B) Hemorrhagic stroke. (C) Ischemic events. (D) Ischemic stroke. (E) All‐cause mortality.
TABLE 2.
Clinical outcomes.
| All cohort | Number (/1000PY) | Crude HR | 95% CI | p value | Adjusted HR | 95% CI | p value |
|---|---|---|---|---|---|---|---|
| Major bleeding | |||||||
| Moderate/Severe anemia | 78 (45.2) | 2.34 | 1.80–3.05 | <.001 | 1.96 | 1.49–2.58 | <.001 |
| Mild anemia | 107 (28.6) | 1.64 | 1.30–2.07 | <.001 | 1.48 | 1.17–1.87 | .001 |
| No anemia | 260 (16.0) | 1.0 | — | — | 1.0 | — | — |
| Hemorrhagic stroke | |||||||
| Moderate/Severe anemia | 5 (2.8) | 0.89 | 0.35–2.27 | .82 | 0.76 | 0.29–1.95 | .56 |
| Mild anemia | 10 (2.6) | 0.95 | 0.47–1.89 | .87 | 0.86 | 0.43–1.74 | .68 |
| No anemia | 42 (2.5) | 1.0 | — | — | 1.0 | — | — |
| Ischemic events | |||||||
| Moderate/Severe anemia | 23 (12.8) | 0.88 | 0.57–1.37 | .67 | 0.63 | 0.39–0.99 | .046 |
| Mild anemia | 60 (15.9) | 1.23 | 0.92–1.66 | .17 | 1.03 | 0.76–1.39 | .87 |
| No anemia | 193 (11.8) | 1.0 | — | — | 1.0 | — | — |
| Ischemic stroke | |||||||
| Moderate/Severe anemia | 21 (11.7) | 0.89 | 0.56–1.41 | .61 | 0.64 | 0.39–1.03 | .07 |
| Mild anemia | 55 (14.5) | 1.24 | 0.91–1.69 | .17 | 1.04 | 0.76–1.43 | .79 |
| No anemia | 176 (10.8) | 1.0 | — | — | 1.0 | — | — |
| All‐cause mortality | |||||||
| Moderate/Severe anemia | 126 (69.5) | 5.07 | 4.01–6.42 | <.001 | 3.77 | 2.95–4.83 | <0.001 |
| Mild anemia | 149 (36.7) | 2.94 | 2.36–3.65 | <.001 | 2.44 | 1.95–3.05 | <.001 |
| No anemia | 207 (12.5) | 1.0 | — | — | 1.0 | — | — |
Abbreviations: CI, confidence interval; HR, hazard ratio; PY, patient‐years.
The cumulative incidences at 4 years of ischemic events in the moderate/severe, mild, and no anemia groups were 7.2%, 6.0%, and 5.2%, respectively (Figure 2C). The incidence of ischemic events with cases per 1000 patient‐years among these three groups was 12.8 versus 15.9 versus 11.8, respectively (Table 2). The adjusted HRs (95% CIs) of moderate/severe and mild anemia for ischemic events were 0.63 (0.39–0.99) and 1.03 (0.76–1.39) compared to the no anemia group (Table 2). The cumulative incidences of ischemic stroke at 4 years among these three groups were 7.0%, 5.6%, and 4.8%, respectively (Figure 2D). The incidence of ischemic stroke with cases per 1000 patient‐years in these three groups was 11.7 versus 14.5 versus 10.8 (Table 2).
The cumulative incidences of all‐cause mortality at 4 years among these three groups were 21.2%, 15.7%, and 5.2%, respectively (Figure 2E). The incidence of all‐cause mortality with cases per 1000 patient‐years in these three groups was 69.5 versus 36.7 versus 12.5 (Table 2). The adjusted HRs (95% CIs) of moderate/severe and mild anemia for all‐cause mortality were 3.77 (2.95–4.83) and 2.44 (1.95–3.05) compared to the no anemia group (Table 2).
The exploratory analyses revealed the adjusted HR (95% CI) of severe anemia for major bleeding was 5.17 (1.99–13.40) and a linear trend from mild to severe anemia was observed (Table S1). The adjusted HR (95% CI) of severe anemia for all‐cause mortality was 5.56 (2.13–14.5) and a linear trend also existed (Table S1). The moderate/severe anemia group had consistently higher HRs in subgroups for major bleeding (Figure S1), and subgroup analyses suggested significant interactions between the severity of anemia and eGFR (interaction p = .04). The sensitivity analyses revealed that the adjusted HRs (95% CIs) of moderate/severe and mild anemia for major bleeding were 2.81 (2.09–3.78) and 1.74 (1.33–2.29) compared to the no anemia in the warfarin alone stratum, but those were 1.22 (0.65–2.27) and 1.24 (0.81–1.90) in the switch to DOACs stratum, respectively (Table S2).
4. DISCUSSION
Analyzing a nationwide registry of consecutive patients in daily clinical practice, we demonstrated the prevalence of anemia and the association between anemic status and bleeding risk in NVAF patients taking OACs. It was found that 28% of NVAF patients taking oral OACs had mild to severe anemia, including 9.6% with moderate or severe anemia. The anemic status was significantly associated with the incidence of major bleeding. The elevated risk of major bleeding was prominent in those with moderate/severe anemia, whose incidence was 14.9% at 4 years, but those with mild anemia were also at higher risk of major bleeding, adjusting for other clinically relevant variables. However, ischemic events or strokes were not associated with the presence or degree of anemia among those patients.
All conditions which necessitated anticoagulants or antiplatelets for preventing ischemic events have a similar dilemma. For example, the prevalence of mild to severe anemia was reported to be 34% among patients with venous thromboembolism (VTE) who started anticoagulants. 14 Among hospitalized Japanese patients with VTE, 54% were reported to have mild to severe anemia. 15 These trends were also observed in acute coronary syndrome patients undergoing percutaneous coronary intervention (PCI) which had a 30% prevalence. 16 The 28% prevalence of mild to severe anemia in our registry was consistent with these reports. Our study shed light on anemia being a common comorbidity in patients for whom anticoagulation is necessary.
It was reported that patients with anemia had significantly higher rates of major bleeding than those without anemia with relative risk of 2.84 among patients with VTE. 14 Among hospitalized Japanese patients with VTE, the HRs for 5 years of moderate/severe and mild anemia relative to no anemia for major bleeding were 1.91 and 1.41, respectively. 15 In our study, the 4‐year HRs were 1.96 and 1.48 for moderate/severe and mild anemia relative to no anemia for major bleeding, respectively. These results were very close to previously reported data and the bleeding risk became higher when anemia became worse in patients requiring OACs.
On the other hand, the risk of ischemic events was not increased by the degree of anemia in our study. The HRs for ischemic events in NVAF patients with moderate/severe and mild anemia to those without anemia were 0.63 and 1.03, respectively. A post hoc analysis of the ARISTOTLE trial, which included over 18,000 patients with AF randomized to warfarin or apixaban showed that anemia was not associated with stroke or systemic embolism (adjusted HR, 0.92; 95% CI, 0.70–1.21). 17 Another study showed that in patients with VTE taking OACs, the rate of DVT recurrences was similar in patients with mild anemia relative to no anemia (relative risk 1.07; 95% CI 0.88–1.30) and the rate of pulmonary embolism (PE) recurrences was somewhat higher in patients with mild anemia but no significant difference was found (relative risk 1.23; 95% CI 1.00–1.56). 14 Considering these observations together, it is considered that anemia does not strongly affect the risk of ischemic events in patients on OACs.
We thus assumed that anemia became a risk stratification variable for bleeding risk in patients with NVAF due to these reasons. The most frequently used risk score for 1‐year major bleeding in AF patients is the HAS‐BLED score, which refers to hypertension, abnormal renal or liver function, history of stroke, history of bleeding events or predisposition of bleeding, labile international normalized ratio, elderly (>65 years), and antiplatelet use or alcohol abuse. 2 In this score, anemia was considered a predisposition of bleeding and the B score was counted if anemia was present. Although bleeding history was well defined as history of any bleeding event requiring hospitalization or causing a decrease in Hb level of >2 g/L or requiring blood transfusion, there was no definition of anemia as predisposition. 2 As a result, anemia was not well recognized as a risk for bleeding events and B score did not reflect the anemic status in calculating the HAS‐BLED score.
In fact, the prevalence of patients with mild to severe anemia but no B score in the assessment of physician in charge was 95% (1992/2100) in our study. Even in patients with the moderate/severe anemia, 93% (676/726) were not considered to have predisposition of bleeding. Our study shed light on there being more patients than expected in daily clinical practice who were at risk for bleeding but whose risk was overlooked, and OACs were started irrespective to the risk of bleeding. Therefore, it is clinically relevant to clarify the impact of anemia severity on bleeding risk among such patients. In the Euro Heart Survey, the odds ratio (OR) for history of major bleeding as a risk factor for major bleeding after taking anticoagulants within 1 year was 7.51 (3.00–18.78). 2 Our exploratory analyses revealed the HR of severe anemia (Hb <8 g/dL) for major bleeding was 6.86 (2.68–17.55) compared to the no anemia group, which was close to the results from the Euro Heart Survey. This category of severe anemia could be a criterion for anemia incorporating the HAS‐BLED score, but further careful validation should be necessary because only 0.3% (23/7558) of the patients in this cohort had severe anemia.
Another important issue is how to deal with NVAF patients with anemia in daily clinical practice. We should investigate the cause of anemia before considering the use of OACs, and several underlying conditions could be corrected before the use of OACs. In addition, if anemia was considered a risk for future major bleeding, blood transfusion before starting OACs could be an option. However, blood transfusion had been reported to increase postoperative mortality and cardiovascular events in patients undergoing PCI. 18 The restrictive red blood cell transfusion was also shown to be non‐inferior to liberal strategies for postoperative mortality, cardiovascular events, or new‐onset renal failure in cardiac surgery patients. 19 Because the pathophysiology was different between ischemic heart disease or structure cardiac disease and NVAF, we should carefully pay attention to the benefit or risk of correction of anemia in NVAF patients even if the anemia itself was a poor prognostic marker.
4.1. Limitations
This study had several limitations. First, the underlying cause of anemia was not identified, and the history of mild bleeding was not examined. These factors could be unadjusted confounding factors. However, most of the included patients were post‐menopausal and the subgroup analysis for malignancy revealed consistent results to the main analyses. We thus considered the effects of these unmeasured factors should be small. Second, all included patients started with warfarin at the start of the cohort and the changes from warfarin to DOACs were determined by the physician in charge. The decision of changes might be influenced by the risk of bleeding and with higher risk of selection bias. We thus conducted the sensitivity analyses which revealed the increased risk of major bleeding in moderate/severe anemia and mild anemia groups were apparent in the warfarin alone stratum, but not apparent in the switch to DOACs stratum. Because 72% of patients continued the warfarin, the finding was not influenced significantly. However, the increased risk of major bleeding due to anemia should be carefully considered in patients on DOACs. Third, we did not collect the details of major bleeding. Because some causes of major bleeding could be associated with anemia, but others were not, the investigation of causes of major bleeding should be carefully considered. Fourth, this study used the baseline Hb at the inclusion of the cohort. Such measurements could be affected by acute conditions such as dehydration and might change over time, leading to misclassification of anemia, but we did not collect the changes in Hb during the hospital stay or other measurements related to dehydration or other conditions. However, such misclassification should occur randomly in the anemia and no anemia groups and the estimates of the effects of anemia should be weakened by the random transitions. Therefore, the estimates with statistical significance should be robust in the large‐scale study. Finally, the patients of this study were selected from participating hospitals, and these findings might not be applicable to patients in other settings. Therefore, the findings of this study should be attested in other settings including patients with different geographic or ethnic backgrounds.
5. CONCLUSIONS
Anemia was frequently encountered in patients with NVAF on OACs in daily clinical practice. The severity of anemia in such patients was associated with a higher risk of major bleeding and mortality but not with the incidence of ischemic events. Our findings were relevant to reduce bleeding complications in patients who were taking OACs. When OACs are considered in patients with NVAF, anemic status should be carefully monitored.
AUTHORS’ CONTRIBUTIONS
Dr. Norito Kinjo carried out study conception and design, statistical analysis, interpretation of data, management, and drafting of the manuscript. Dr. Shinichiro Ueda was involved in study conception and design, data acquisition, and critical revision of the manuscript. Dr. Kazutaka Uchida carried out statistical analysis, interpretation of data, and critical revision of the manuscript. Dr. Fumihiro Sakakibara, Dr. Mari Nezu, and Dr. Hideki Arai were involved in interpretation of data and critical revision of the manuscript. Dr. Takeshi Morimoto: study conception and design, data acquisition and management, interpretation of data, management, and drafting of the manuscript. All authors: final approval of the manuscript.
FUNDING INFORMATION
This study was supported by Bristol‐Myers Squibb. The funders did not participate in any part of the study, from conception to article preparation.
CONFLICT OF INTEREST STATEMENT
Dr. Kinjo, Dr. Sakakibara, Dr. Nezu, and Dr. Arai have no disclosures to report. Dr. Uchida reports lecturer's fees from Daiichi Sankyo. Dr. Ueda reports receiving a research grant from Bristol‐Myers Squibb, Chugai, Kowa, MSD, Pfizer, and Takeda, lecturer's fee from Boehringer Ingelheim, MSD, and Taiho, and manuscript fees from Kowa. He served on an advisory board for Otsuka.
Dr. Morimoto reports lecturer's fees from Bayer, Daiichi Sankyo, Japan Lifeline, Kyocera, Mitsubishi Tanabe, Novartis, and Toray; the manuscript fees from Bristol‐Myers Squibb and Kowa; served advisory boards for Asahi Kasei, Boston Scientific, Bristol‐Myers Squibb, and Sanofi.
ETHICS APPROVAL STATEMENT
The institutional review boards of the University of the Ryukyus (No. 597) and all 71 participating centers approved the study protocol (Supplemental Table). The institutional review boards waived the need for written informed consent and approved the study in accordance with the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan.
PATIENT CONSENT STATEMENT
Not Applicable.
CLINICAL TRIAL REGISTRATION
Not Applicable.
Supporting information
Figure S1.
Tables S1–S2.
ACKNOWLEDGMENTS
We are indebted to the data managers of the Institute for Clinical Effectiveness (Ms. Makiko Ohtorii, Ms. Ai Sunagawa, Ms. Kaori Yamamoto, Ms. Sachiko Kitamura, Ms. Hirono Saito, and Ms. Saeko Nagano) for managing the data and performing the statistical analyses.
Kinjo N, Ueda S, Uchida K, Sakakibara F, Nezu M, Arai H, et al. Impact of anemia on major bleeding in patients taking oral anticoagulants for nonvalvular atrial fibrillation. J Arrhythmia. 2023;39:556–565. 10.1002/joa3.12885
REFERENCES
- 1. Culebras A, Messé SR. Summary of evidence‐based guideline update: prevention of stroke in nonvalvular atrial fibrillation: report of the guideline development Subcommittee of the American Academy of neurology. Neurology. 2014;83:1220. 10.1212/wnl.0000000000000831 [DOI] [PubMed] [Google Scholar]
- 2. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user‐friendly score (HAS‐BLED) to assess 1‐year risk of major bleeding in patients with atrial fibrillation: the euro heart survey. Chest. 2010;138:1093–100. 10.1378/chest.10-0134 [DOI] [PubMed] [Google Scholar]
- 3. Chang SH, Chou IJ, Yeh YH, Chiou MJ, Wen MS, Kuo CT, et al. Association between use of non‐vitamin K Oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation. JAMA. 2017;318:1250–9. 10.1001/jama.2017.13883 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. O'Neal WT, Sandesara PB, Claxton JS, MacLehose RF, Chen LY, Bengtson LGS, et al. Influence of sociodemographic factors and provider specialty on anticoagulation prescription fill patterns and outcomes in atrial fibrillation. Am J Cardiol. 2018;122:388–94. 10.1016/j.amjcard.2018.04.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Bach V, Schruckmayer G, Sam I, Kemmler G, Stauder R. Prevalence and possible causes of anemia in the elderly: a cross‐sectional analysis of a large European university hospital cohort. Clin Interv Aging. 2014;9:1187–96. 10.2147/cia.S61125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:1139–51. 10.1056/NEJMoa0905561 [DOI] [PubMed] [Google Scholar]
- 7. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981–92. 10.1056/NEJMoa1107039 [DOI] [PubMed] [Google Scholar]
- 8. Morimoto T, Uchida K, Sakakibara F, Kinjo N, Ueda S. Effect of concomitant antiplatelet therapy on ischemic and hemorrhagic events in patients taking oral anticoagulants for nonvalvular atrial fibrillation in daily clinical practice. Pharmacoepidemiol Drug Saf. 2021;30:1321–31. 10.1002/pds.5228 [DOI] [PubMed] [Google Scholar]
- 9. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of atrial fibrillation. JAMA. 2001;285:2864–70. 10.1001/jama.285.22.2864 [DOI] [PubMed] [Google Scholar]
- 10. Nutritional anaemias . Report of a WHO scientific group. World Health Organ Tech Rep Ser. 1968;405:5–37. [PubMed] [Google Scholar]
- 11. Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non‐surgical patients. J Thromb Haemost. 2005;3:692–4. 10.1111/j.1538-7836.2005.01204.x [DOI] [PubMed] [Google Scholar]
- 12. Butzer JF, Cancilla PA, Cornell SH. Computerized axial tomography of intracerebral hematoma. A clinical and neuropathological study. Arch Neurol. 1976;33:206–14. 10.1001/archneur.1976.00500030062014 [DOI] [PubMed] [Google Scholar]
- 13. Hicks KA, Tcheng JE, Bozkurt B, Chaitman BR, Cutlip DE, Farb A, et al. 2014 ACC/AHA key data elements and definitions for cardiovascular endpoint events in clinical trials: a report of the American College of Cardiology/American Heart Association task force on clinical data standards (writing committee to develop cardiovascular endpoints data standards). J Am Coll Cardiol. 2015;66:403–69. 10.1016/j.jacc.2014.12.018 [DOI] [PubMed] [Google Scholar]
- 14. Kuperman A, López‐Reyes R, Bosco LJ, Lorenzo A, José B, Farge Bancel D, et al. Anemia and bleeding in patients receiving anticoagulant therapy for venous thromboembolism. J Thromb Thrombolysis. 2018;45:360–8. 10.1007/s11239-018-1610-9 [DOI] [PubMed] [Google Scholar]
- 15. Yamashita Y, Morimoto T, Amano H, Takase T, Hiramori S, Kim K, et al. Influence of baseline anemia on long‐term clinical outcomes in patients with venous thromboembolism: from the COMMAND VTE registry. J Thromb Thrombolysis. 2019;47:444–53. 10.1007/s11239-018-1791-2 [DOI] [PubMed] [Google Scholar]
- 16. Nikolsky E, Mehran R, Dangas G, Fahy M, Na Y, Pocock SJ, et al. Development and validation of a prognostic risk score for major bleeding in patients undergoing percutaneous coronary intervention via the femoral approach. Eur Heart J. 2007;28:1936–45. 10.1093/eurheartj/ehm194 [DOI] [PubMed] [Google Scholar]
- 17. Westenbrink BD, Alings M, Granger CB, Alexander JH, Lopes RD, Hylek EM, et al. Anemia is associated with bleeding and mortality, but not stroke, in patients with atrial fibrillation: insights from the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial. Am Heart J. 2017;185:140–9. 10.1016/j.ahj.2016.12.008 [DOI] [PubMed] [Google Scholar]
- 18. Kwok CS, Sherwood MW, Watson SM, Nasir SB, Sperrin M, Nolan J, et al. Blood transfusion after percutaneous coronary intervention and risk of subsequent adverse outcomes: a systematic review and meta‐analysis. JACC Cardiovasc Interv. 2015;8:436–46. 10.1016/j.jcin.2014.09.026 [DOI] [PubMed] [Google Scholar]
- 19. Mazer CD, Whitlock RP, Fergusson DA, Hall J, Belley‐Cote E, Connolly K, et al. Restrictive or Liberal red‐cell transfusion for cardiac surgery. N Engl J Med. 2017;377:2133–44. 10.1056/NEJMoa1711818 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1.
Tables S1–S2.